Wrapper function for Scrublet doublet detection parameters

Constructor for the various Scrublet parameters. In this case, the default for the kNN graph generation was set to "hnsw" as this algorithm showed the best performance in different empirical benchmarks.

Usage

params_scrublet(
  sim_doublet_ratio = 1.5,
  expected_doublet_rate = 0.1,
  stdev_doublet_rate = 0.02,
  n_bins = 100L,
  manual_threshold = NULL,
  normalisation = list(),
  hvg = list(),
  pca = list(),
  knn = list(k = 0L, ann_dist = "euclidean")
)

Arguments

sim_doublet_ratio: Numeric. Number of doublets to simulate relative to the number of observed cells. For example, 2.0 simulates twice as many doublets as there are cells. Defaults to 1.5.
expected_doublet_rate: Numeric. Expected doublet rate for the experiment, typically 0.05-0.10 depending on cell loading. Must be between 0 and 1. Defaults to 0.1.
stdev_doublet_rate: Numeric. Uncertainty in the expected doublet rate. Defaults to 0.02.
n_bins: Integer. Number of bins for histogram-based automatic threshold detection. Typically 50-100. Defaults to 100L.
manual_threshold: Optional numeric. Manual doublet score threshold. If NULL (default), threshold is automatically detected from simulated doublet score distribution.
normalisation: List. Optional overrides for normalisation parameters. See params_norm_doublet_detection_defaults() for available parameters: log_transform, mean_center, normalise_variance, target_size.
hvg: List. Optional overrides for highly variable gene selection parameters. See params_hvg_defaults() for available parameters: min_gene_var_pctl, hvg_method, loess_span, clip_max.
pca: List. Optional overrides for PCA parameters. See params_pca_defaults() for available parameters: no_pcs, random_svd, sparse and skip_first_pc.
knn: List. Optional overrides for kNN parameters. See params_knn_defaults() for available parameters: k, knn_method, ann_dist, search_budget, n_trees, delta, diversify_prob, ef_budget, m, ef_construction, ef_search, n_list and n_probe. Note: this function defaults to k = 0L (automatic neighbour detection).

Value

A named list with all Scrublet parameters, combining defaults with any user-specified overrides.